Identification and Inhibiting Mechanism of Safety Violations of Electric Bicycle Riders Based on Machine Learning

Abstract

With the rapid rise of electric bicycles as a green mode of transportation, road traffic safety issues associated with their use have become increasingly severe. In response to the safety violations of riders, this study constructed a comprehensive classification system covering 21 static and dynamic risk factors, and integrated open street view data, self-collected images, and social media data, totaling 6193 samples. The study adopted a graph neural network model (GAT) based on the attention mechanism and an improved PLCJ algorithm to achieve efficient identification of violations during riding and risk factor correlation analysis, and the classification accuracy was significantly better than the traditional method. Technically, we leverage anomaly-detection concepts from the Internet of Vehicles (IoV) and integrate graph embedding over multisource data. We then apply this to behavior modeling and dynamic violation detection in nonmotorized scenarios. This approach extends IoV safety analysis into micro-transportation systems. Experimental results show that the PLCJ algorithm significantly outperforms traditional methods in classification accuracy on both medium- and large-scale datasets. The GAT model adaptively assigns weights, allowing for precise identification of core risk factor combinations linked to various electric bicycle violations. Based on these findings, the study proposes multidimensional management strategies: optimizing road design to reduce conflicts between motorized and nonmotorized traffic, advancing intelligent monitoring technologies, implementing targeted safety education initiatives, and enhancing traffic resource allocation through a global attention model (GA-GNN). This study not only provides theoretical support for urban traffic safety governance but also opens up a new direction for the application of machine learning algorithms in non-motor vehicle behavior anomaly detection in the IoV environment, helping to build a more intelligent, safe, and sustainable urban travel system.